Cyberattacks have become a global problem, the magnitude of which threatens the stability of governments, businesses, and critical infrastructure. Water distribution systems are targets of cyberattacks with serious consequences for millions of users. This risk has increased due to the rise in digitization and connectivity. Thus, the main objective of this research is to detect cyberattacks in water distribution systems using Hybrid Machine Learning Voting. The proposed methodology consists of four phases: obtaining the dataset (BATADAL); preprocessing (filling in null values, generating temporal variables, normalization, standardization, and balancing with SMOTEENN); implementing the models (Random Forest, XGBoost, SVM, CatBoost, Hybrid Machine Learning Voting, LSTM and CNN); and evaluating the models. The optimal results were obtained with the SVM algorithm, whose metrics were superior to the other models with the metrics Accuracy, Precision, Recall, F1-Score. AUC and Threshold, whose results were 0.9938, 0.9828, 0.9853, 0.9840, 0.991, and 0.90. In conclusion, the results demonstrate that the proper implementation of hybrid and machine learning algorithms is efficient for detecting and predicting cyberattacks in water distribution systems.